Pattern analysis and performance accounting for agile code development may be provided. Conventional large scale, highly used service development often faces the challenge of decreasing errors and increasing the speed of code flow from developer to production. Traditional testing approaches identify the performance of new code pieces or components by relying on heavy infrastructures where suites of tests are run to identify performance regressions. In some situations, the time from when the code is checked in by a developer to the time tests are run may be pretty significant. Conventional test methodologies also do not provide instant feedback to the developer as to whether the code will use less or more resources (e.g., processor usage, disk operations, memory, network, etc.) and/or if performance in terms of latencies or execution times will improve or deteriorate.